Efficient Relay Selection and Transmission Scheduling in Non-Terrestrial Assisted Vehicular Networks

Abstract

Millimeter-wave communication provides high capacity channels suitable for vehicular networks, yet it suffers from susceptibility to dynamic environmental blockages caused by urban infrastructure, terrain, and mobility. To mitigate these limitations, we introduce an innovative two-way relay architecture supported by unmanned aerial vehicles, aimed at improving connectivity and expanding coverage for moving vehicles. Additionally, in scenarios where direct vehicle-To-vehicle links or unmanned aerial vehicle-Assisted relaying are unavailable due to severe obstructions or coverage gaps, the system seamlessly falls back on satellite-To-vehicle communication to maintain reliable data delivery. The overall scheduling process is modeled as a joint optimization problem combining relay selection and transmission scheduling, with the goal of minimizing end-To-end communication latency while satisfying throughput demands. To solve this problem, we explore a range of techniques including constraint programming, deep Q-learning, and proximal policy optimization. Simulation results show that the proposed PPO based approach achieves near-optimal performance, with up to 7.5% higher throughput and 6.27% fewer time slots than JRDS across various deployment scenarios. © 2025 IEEE.